-1 to Brian for the double post. :-P
“The player development is the main thing I like (about Michigan),” Williams said. “You can see that they develop their players. They get them in the gym and they work them hard. And their hard work pays off.”
Annual exercise. This site, like all other sites, has a favored playoff proposal that it spent way too much time justifying a couple years ago and wouldn't like to revisit in depth but would like to remind you how much better it would be than the current broken system.
In short, it's a six-team playoff with home games in the first two rounds and the final at the Rose Bowl. There are no autobids. First round would be the weekend after the Championship games (ie, this weekend), the second round would be January 1st, and the final would be the 8th or whatever in the Rose Bowl. This year's edition:
1. Alabama vs. 4. TCU / 5. Boise State
2. Texas vs. 3. Cincinnati / 6. Florida
That is weirdly like two of the existing BCS games, but this time they actually, you know, get a shot at the big boys should they win.
A brief recap of why this is good:
Yea, truly this is the way.
The most evil corporation in the history of world. …probably isn't this one I'm about to talk about. But they are irritating.
No one else probably remembers this, but when I saw the letters in this Sporting Blog post I was reminded of a previous fiasco:
Back in late October, SI broke the story of a fight for film going on between the NFL and XOS Technologies. As a quick refresher, XOS is in charge of digitizing film for eight conferences, including the SEC, Pac-10, Big 12, Mid-American, WAC and Sun Belt. Now XOS wants what is reportedly $20-$30 million for film the NFL used to receive for nothing.
You may remember XOS from such ill-fated public relations disasters as "The SEC Controls Your Mind"; they signed on to do all the digital stuff for the SEC this year and immediately issued a draconian policy that was RIAA-esque in its blinkered belief that you can control the internet. Now this same corporation is trying to extract money from the NFL for tapes of prospects, and the NFL is in a fight. I bet coaches who want to tout the NFL-readiness of their players just love this corporation.
You'll note the Big Ten is not afflicted; they've cast their lot with Fox and the Big Ten Network and should have no problems getting seniors and—more importantly for folks hoping Donovan Warren reconsiders—juniors scouted in a timely fashion.
Aigh. Formerly Anonymous ran across this interview with Richard Billingsley, which gives me another opportunity to rail against his hodge-podge of dreck that purports to be a computer rating system. It's not a surprise that a guy who's described his system like this…
Believe it or not, the system is designed after our own United States Constitution. But don't hold that against it! Although at times I feel this system is just about as complicated as our Federal Government, there is one huge difference..... this one works!
…"doesn't even have a degree," according to the article, and won't divulge his formulas for ranking teams. Because they're insane. Things Billingsley does in his rankings:
This is an organization that thought a propaganda website, widely-mocked twitter account, and Ari Fleischer were good ideas, so it's not a huge surprise that they've cast their lot with a guy who sounds like he owns a bunch of stuff from the Franklin Mint. It is further confirmation that the people in charge of college football are sort of clueless: they've cast their lot with an aging nut's secret blend of herbs and spices that hasn't been updated since 1970.
PREWB! This was discussed extensively on the radio but has gone uncommented on here: the latest Free Press bit on the Michigan State "fight" is further evidence that the paper is in the tank for the Spartans. It interviews two parents of suspended players but doesn't bother to talk to the very talkative victim, mention that the president of the fraternity in question would like to see the football players gone for good, or mention anything in Glenn Winston's past. It is strictly from the POV of State players' parents. Jon Chait has a fuller takedown of an organization that's descended into self-parody.
This quote from the "grand polaris" of Iota Phi Theta—what a title—is from the News…
"From what I understand, almost all of those guys, if not all of them that came in there, (threw) a punch," Price said of the eight who received suspensions. "They came in, (based) on their behavior, with the sole purpose of beating up or physically abusing an Iota."
…and is followed up by the term "assault," not "fight." The News also reports on Dantonio's response to the thing. Like the context left out of the Free Press Katrina story and provided by AnnArbor.com, you don't have to take a crazy biased blogger's word for it. You can observe the other news organizations covering the story and contrast them with the Free Press.
Etc.: Chengelis on the AD search. Harbaugh is not talking to Notre Dame. Bill Taylor discusses his company, which deals with substance abuse. Red is 70, and it's been suggested that Happy Birthday should be sung on Friday at Yost.
-1 to Brian for the double post. :-P
Not just for the plebes anymore!
... so it's not a huge surprise that they've cast their lot with a guy who sounds like he owns a bunch of stuff from the Franklin Mint.
It's OK, though, I HAVE A SPECIAL OFFER FOR YOU:
you get an "unofficially" signed Obama Inaugural Pendant.
I'd rather see a 16-team playoff
there is now way you can justify leaving Oregon out. The P10 was easily the 2nd best conference this year, and I don't care if Oregon lost to Boise.
like you, I came up with an 8 team format about 5 years ago. No serious contender would be left out, no home games (due to climate), play would start this week and conclude by early January (not much missed classes), only 4 teams would be playing an extra game (because they all would have been in a bowl game anyway) and the championship gets decided on the field. Some of the key bowls could opt to host the semis and final on a rotating basis, or keep what they have now.
In any case, a playoff is needed and there are no acceptable excuses for not having one. All of the other sports manage to have them (including the other football divisions).
Is acceptable reason to not.
How do you get anyone to go to the games if they're not at home?
And due to climate? What, are we competing in rhythmic gymnastics now?
My reasoning is that I don't want any system that potentially compromises the importance of OSU-UM
or shouldn't be. Doesn't matter if both teams are 12-0, both teams are 6-6, qualified for bowls, qualified for BCS, whatever.
UM-OSU is important no matter what happens afterward. If we have the chance to knock them down or out of a postseason system, so much the better, but still ... the game will always be important unless you are one of those fans or players that needs something to get you fired up for that game, and in that case, I don't know what to tell you.
Unless you think that 9-4 Virginia Tech belonged in the playoff ahead of three of 12-0 Utah, 12-0 Boise State, 12-1 Alabama, 11-1 Texas, and 11-1 Texas Tech last year, that's an idea that sounds great but doesn't actually work.
No auto-bids for conference champions unless you do a 16-team bracket with all 11 champs. If you insist on rewarding conference championships (which is reasonable), make a stipulation that at most two teams who did not win their conferences can be put in - but do NOT restrict which conferences those other six won.
Going by the BCS standings as a proxy for the selection committee, here are the instances in which a non-BCS team would have qualified:
This year: #4 TCU and #6 Boise State ahead of #8 Ohio State (who qualifies as an at-large anyway) and #9 Georgia Tech (who does not)
Last year: #6 Utah and #9 Boise State ahead of #12 Cincinnati (11-2) and #19 Virginia Tech (9-4)
'07: Nobody (Hawaii was behind all six BCS conference champions plus Georgia and Missouri)
'06: #8 Boise State (and the fifth-highest conference champ ahead of Oklahoma) ahead of #14 Wake Forest (11-2)
'05: Nobody (although #6 Notre Dame would be out due to not being in a conference to get a championship; if you consider them "champion of the independents" it bumps West Virginia out)
'04: Utah as #6 (and the fourth-highest conference champion!) and Boise State (ranked #9) go ahead of #13 Michigan (9-2) and #21 Pitt (8-3). (#10 Louisville, in CUSA at the time, also gets passed over.)
'03: Nobody (Fake Miami is the seventh-highest conference champ)
'02: Nobody (the last spot comes down to #9 Notre Dame or #14 Florida State, depending on the same interpretation as '05)
'01: Nobody (none even in the BCS top 15)
'00: Either Notre Dame (at #11) or TCU (at #14) gets the last spot ahead of three-loss Purdue.
'99: Marshall (at #12) takes the last spot over the Pac-10 champion, three-loss Stanford.
'98: Tulane gets the nod ahead of three-loss Syracuse.
The only issue with Brian's proposal is that it would de-value the two BCS bowls that are essentially left out of the playoff each year.
Personally, I think it would be a small price to pay for an equitable playoff format, but I don't think the two bowls would agree. Maybe if they all agree to rev share from all the playoff games, including the two home games.
BTW I'd love making one of the southern schools come up and play us at the Big House in December.
Where's jamiemac? We need odds on the likelihood of any of the Internet's Brilliantly Perfect Playoff Systems being the system the NCAA institutes, if/when it does.
The second sign is boasting at a sub-literate level about phantom sexual conquests at Skeeps.
The third sign is POSTING IN ALL CAPS. YES.
The fourth, and fatal sign, is concluding that Michael Rosenberg is awesome.
I won't be too concerned until Brian begins concluding every post with:
I wouldn't give that story one Iota of credence. Ugh. Someone else can do better than that.
I agree 100% that the Freep is tilting the Potluck Massacre story toward daylight for MSU. It was no more a "fight" than date rape is "consensual sex". The Freep also used the term "fracas" (defined as a "noisy quarrel").
That being said, I don't buy that they're out to get UofM and are pro-MSU, but the end result in the paper sure feels that way. My best guess is that the Newspaper industry lost its best talent...so journalistic integrity falls prey to laziness.
I hardly ever comment on Brian's post because what am I going to say? Wow...I agree with you. Like always. But this one seems relevant.
I have been writing my Advanced composition final, a research paper about a playoff and found out I was allowed to cite MGoBlog. I found Brian's playoff idea from 2006 and used it for my paper. While being distracted I checked up on mgoblog like usual throughout the day and found this post. Crazy how things work out I guess.
Once again Brian fails to acknowledge that the initial story itself (ie. Everyone was fighting, players wore ski masks, there were serious injuries, women were assaulted, etc) was based solely on comments from the fraternity and it's members with the vast majority of the info coming from ONE fraternity member and his lawyer.
So I would contend that the Free Press did "bother to talk with the victim" because the entire narrative of the initial story was formed from this source.
The victim's side of the story has somehow managed to influence Brian's thoughts on the matter despite the bias and collusion of the Free Press. :rolleyes:
Yeah, stick it to the victims!
obviously the initial story came from quotes from the victims, the ski-masked perps had not yet been identified. The freep should be trying to get to the truth, not just present the other side of the story.
I agree. For 10 days the only side being reported was from the fraternity, nearly all from 1 particular fraternity member who had already retained legal council and was threatening a lawsuit. Take from that what you will.
After 10 days one sided reporting (mainly because no one on the MSU side was permitted to talk), information from the players' perspective was leaked through side channels (their parents). This was presented as evidence of biased reporting. I find that claim extraordinarily bizarre without a hefty dose of bias on Brian's (and Chait's) part.
The investigation is still ongoing and there is a lot we still don't know yet. The truth could very well be similar to the initial story. A gang of football players wearing ski masks assaulted multiple people seriously including several women. Or, it may be revealed that only one player hit anyone, the injuries were not serious, there were no ski masks, no women were harmed and the incident was sparked by the victim assaulting a football player the night before as the players' parents suggest.
I'm assuming the truth lies somewhere in the middle. Isn't that a more reasonable position than what has been promoted on this blog? Both in regard to the MSU story and allegations of bias on the part of the Free Press?
of MGoBlog (and the Wolverine) on one side and The Detroit Free Press on the other is the exact point of Brian's argument. The Freep leaves no stone unturned in its on-going investigations of the Michigan football program; it behaves differently when dealing with Michigan State football program.
That is not my positioning at all. The Free Press reported one side of the story (negative MSU) for 10 days before reporting the other side. By then the widely accepted narrative of a ski mask wearing posse of women beaters was already established as evidenced by this blog and the comments contained therein. That is pro-MSU bias?
Since everyone else is doing it, here's my take.
11 teams. If you're going to recognize a conference as an FBS conference, their champion gets in. Each champion gets in, and no one else does. That will go a long way towards keeping a serious value on the regular season.
First two rounds are at home field of the higher seeded team, semis and finals are at current BCS bowl sites. Obviously one gets left out a year, that will have to be figured out. And Notre Dame would just be hosed I guess. Time to join up. Here's how it would look this year.
EDIT: I am not good at programming the internets, just click on it to make it bigger.
Very simple. Return to the four traditional bowl matchups, making sure a spot is reserved for a non-BCS school. This leaves 6 conference champs, one at large team (this year Florida would be it) and another at large non-BCS school (TCU, sorry Boise). The four winners play the next week, and the championship game is played the weekend before the Superbowl. Problem solved.
1) Travel. You really think you're going to get fans to fill the stadiums three weeks in a row on short notice? (Basketball manages it, but they have four or eight teams to fill a 20,000 seat arena, not two teams to fill a 90,000 seat one. And even then the first-weekend games are sparsely attended.)
2) Same issue as I mentioned to someone else who proposed an eight-team playoff with six auto-bids: In no way does a 9-4 ACC champion (or Nebraska this year, had they beaten Texas) belong ahead of a 12-0 MWC or WAC team. If you cap the number of automatic bids at something less than 11, you cannot specify which conferences get them. Take the six highest-ranked conference champions; if that happens to be Big 10, CUSA, MWC, WAC, MAC, and Sun Belt (don't ask me how that would happen), so be it.
3) Sticking to the "traditional" matchups is too likely to create a 1v2 quarterfinal. Take a hypothetical year where Michigan and OSU are both highly ranked to the point where the loser is probably the at-large team, and USC is ranked #1. The loser of Michigan-OSU gets an easier road through the tournament than the winner.
And the reason 90% of most playoff systems fail. No one who's not a fan of the team is going to want to see ONE game of someone else's team play an early round game. It's not like B-Ball where you can stage 5 games at a site (or is used to playing in front of 3/4 empty stadiums). And no one is going to book to trips just to GET to the championship game...they're going to wait to see if their team gets in.
That's why when I heard it, I really like his system. I'm old school pro bowl, but the BCS is worse than when the Champs all went to separate bowls and the polls just did it. So this takes care of how to get people to the games. It keeps the bowls valuable, only really taking out the four (2 of which are already yanked). It doesn't take too many teams, so the best regular season still has value. It doesn't take so few that you're not opening yourself up to complaining - the #7 team doesn't have a lot of likelihood of being the best team...far less than #5 or #3...and yes, lower teams have won in the NFL playoff, but that's their flaw...that the best team doesn't always win. They're the Champs but you'll never convince me that the 18-1 Patriots were worse than the Giants as a team just because when they split their season series 1-1, they lost the 2nd game.
On your #1, I think College Football is so popular at this point that you can fill the stadiums three times in a row, even if it's not the same group of people. On your #2, (and this is highly controversial but it's my view) I don't care if some of the minor conferences get disadvantaged. I don't think the conferences you mentioned play equivalent football. My bigger problem with your proposal is that there is just no way that there are more than 1-3 teams from non BCS conferences that can compete in any year, with the BCS conference runner-ups, let alone champions. If you put 11 teams in, you instantly promote say the MAC champion over a 2nd place team from a BCS conference that would DESTROY the MAC champion in any game. That's why IMO you have to keep it to the BCS conference champs and reserve a spot for the inevitable minor conference champs that does have a chance to win.
"I think College Football is so popular at this point that you can fill the stadiums three times in a row, even if it's not the same group of people."
You might be able to if the games are at home, but you can't get the bowls to sell out with just one game and two fan bases when you have to travel too.
On the one hand, basketball is a bad example, because sites are determined in advance, so you have about a year to purchase tickets. The assumption is that a lot of fans buy tickets regardless of the teams playing; this should be true to varying but lesser extents in football. (Kind of like how some bowl games sell out and others, no.)
First-weekend games may be sparsely attended, but they do sell very well. In the 20 years prior to 2009 (the NCAA hasn't posted the 2010 record book yet, so no figures on 2009's tournament easily available), as a whole, first-session games sold to no less than 80% of capacity, and in 16 of 20 years, they sold to no less than 85%. (The regionals, surprisingly, seem to sell about the same percentage, except in 2008 when they sold over 100% of capacity.)
I think it would work like basketball does: some stadiums would sell as many tickets as they could in a week and others wouldn't sell out a year in advance. That's one reason I suspect that whatever season is eventually implemented will retain some bowls as playoff sites ... for example, 16 teams, first round at campus sites, remaining games at predetermined sites.
Another thing about attendance: some/a lot/most tickets are marketed to companies, not "real fans", and companies buy that stuff in advance no matter who's playing. (I put real fans in quotes because a) there are some real fans at companies and b) I got tickets to the Midwest Regionals last year, so at least someone was using the tickets wisely, even if it meant I was kind of rooting for Sparty - no offense meant. I mean, Louisville?)
Excellent point about selection (which is why I believe a 16-team playoff will be the eventual winner). One of the ideas is to include enough teams that anyone with a legitimate claim is in. A 4-loss Big Ten champ ahead of a 1-loss SEC runner-up ... okay, that would be funny, but really isn't what we want to see.
Basketball fills stadiums of 1/2 to 1/4 capacity, not trying to fill 80-100,000 seat stadiums. And they're not held in the same place every year, so when it happens, it becomes a regional event. Events in the same place year after year (like bowls) rarely sell out. Rose Bowls regularly play under capacity (especially if the teams have been there recently), and lesser bowls have wide spaces of empty seats. And there will be even more if the fans of the two teams aren't traveling because they're waiting to play in the BIG game.
And your basketball stats being recent reflect only the newer pod system that tries to get teams to okay closer to home to sell more tickets. Back when onlybthe top seeds got any consideration and everyone else was placed willy-nilly across the country would have been a better comparison, for what it's worth.
"Byes and home games keep huge tension in the regular season. The SEC championship game didn't eliminate Florida but instead of a bye and a second-round home game, Florida gets to go to Cincinnati in a quarter-final."
Playing for a bye is not NEARLY the same as an elimination game. You can argue it won't KILL the regular season to NBA/NHL/NFL levels, but it will certainly take a great deal away from these kind of matchups.
Any system that compromises games like this (e.g. '06 UM vs OSU game) should at least acknowledge it as a flaw in the proposal.
Personally, I'd like to see a playoff only if there are auto-bids from each of the 6 BCS conferences. Add in two selection-committee style for non-BCS teams. Puts the focus back on the traditional win-the-conference oriented goals that existed during the Bo-Woody.
"win-the-conference oriented goals"
Yes pls. I hate this obnoxious national championship foofaraw. Conference champions are now and will always be the least disputable champions in college football, and I wish that goal would be recognized as the most important.
Until the NCAA overtakes the conferences and puts together an NFL-style centrally-organized football league with a more balanced, less regionally compartmentalized, a playoff system won't actually make D-1A college football better. It'll make it different, but not better. It'll just have different problems.
Play a 6-way championship game on a hexagonal field.
an unusual approach but...
..."circle jerk" is the exact phrase i think of every time i read someone's proposal for a college football playoff, including brian's.
The Big Ten, ACC and Pac 10 would have serious issues with a 6-team playoff that includes Cincy, TCU and Boise, but none of their schools. I think the baseline for a playoff would be to give automatic bids to each BCS conference champ. You'd probably have to have a couple open spots to give the non-BCS schools a chance to make the playoff, so a 8-team playoff is more realistic.
Too much money in the bowl games.
Our best hope is a "plus one" game. That way the bowls would be the same except two bcs bowls (maybe a yearly rotation) would be 1 v. 4 and 2 v. 3 and then another rotating bowl would host winners of the previous matchup.
This method could keep the amount of bowl games the same, it is the only "playoff" system I could imagine feasibly happening.
of the championship picture.
This year it would be Boise, but better to have one team left than 3 like there are this season.
Someday there will be a playoff, and we'll all wonder why it didn't happen long ago.
What about a college football playoff that:
Keeps All Current Bowl Games In Place
Keeps Most Traditional Rivalries in Place for Bowl Games
Includes 16 Teams
Limits Additional Games
Reduces the Need/Desire for Teams to Schedule “Non-Competitive” Games
Impossible? Not really, here it is:
All 6 BCS conferences will have two divisions (ACC, Big 12, and SEC already have this in place). Big East, Big 10, and Pac 10 will have the option to add teams and split into 2 divisions. If Big East, Big 10, and/or Pac 10 do not want to split into divisions, there would be more “at large” teams. All conferences may add teams if they want up to a maximum of 10 teams per division.
Schedule will consist of 12 regular season games. Teams may NOT play more than 3 non-conference games.
Playoff will consist of 16 teams
The 12 Division winners automatically qualify.
There will be an additional 4 “at-large” bids determined by BCS ranking.
First round of the playoffs (16 teams) will be the same as the current Conference championships (played the week after the regular season ends) plus playoffs between the 4 “at-large” teams. BCS ranking will determine the home team.
Second round of the playoffs (8 teams) will be the following week. Home team determined the same as for the first round.
Third round of the playoffs (4 teams) will be on New Year’s Day at two of the existing bowl games.
Championship game will be the next week at an existing bowl game (like it is now).
All teams that do not make the final 4 are eligible to play in any of the other bowl games (just like now).
All bowl games can keep traditional rivalries except for the 2 bowls hosting the final 4.
Of all of the proposals out there, I like the following:
1. We use the format as it currently exists, with elimination of the Championship Game. Autobids remain as they currently do, as to criteria and selection process or at large.
2. Winners of the Rose, Orange Suger and Little Caesar's Bowls (ok, I guess we could use Fiesta instead of the Pizza - either way, whatever) advance to a "semi-final", with winner of Rose playing winner of Orange and winner of Suger playing winner of Fiesta. The locations of these two games could rotate between these four locations.
3. Winners of the semi-final games advance and meet in an actual National Championship game, to be held in one of the four locations mentioned above.
This would extend the season by 1 extra game for only four teams, and it would extend the season by 2 games for only 2 teams. If the 4 Bowls were played on New Years Day (as they should be), with the two playoff games being played 1 week later (can be 8 or 9 days, depending on what day new years falls), this would amount to the NC game being played approximately 1 week later than it currently is (this year it is on the 7th). This would also preserve the importance of the regular season, maintain the integrity and tradition of the bowls, not diminish the importance of winning the conference and would result in additional TV. Also, even in a year such as this, where there are 5-6 teams that each have a legitimate claim to play for the NC, all would have their chance.
So simple, yet so perfect.
is it hurts the existing bowls to much. Fans aren't going to travel to a bowl game when they'll be in even bigger bowls in the semi final and final.
If you want an 8 team or bigger playoff you might as well do away with the big bowls.
That's a good point - hadnt't thought about that. Using Michigan as an example, while I might be able to swing 1 trip to Pasadena with the family (from NY), I certainly couldn't go out there twice or 3 times. I guess one question is whether there are enough fans that would travel for us to use our team's allocation of tickets for all three games. For Michigan, I think that the answer is yes, because of the size of our fanbase (and the fact that our fans are spread around the country). For a team like Boise or TCU, this might pose a problem, though. Damn you for so quickly and easily destroying the system that I have espoused for about 3 years now!!
It's everybody. Why would you pay to go to see Michigan go to the Orange Bowl, when in two weeks you can pay to see them at the Rose for all the marbles?
Having more fans doesn't mean enough to fill each bowl...just more competition for the final game. There's a reason most sports play playoff games at home till the championship.
I agree that the Billingsly poll is deeply flawed. But your post made me think about the SOS issue: namely, whether to credit the team with the opponent rating at the time of the game or the end of the season. Clearly, the latter is a more reliable indicant of opponent quality. Yet, teams change over the course of a season, sometimes getting better as young players develop or getting worse when players are injured. So, I wondered: is there a tradeoff between the reliability and timeliness of the opponent rating?
But Billingsly doesn't really solve this problem.
Consider the OSU home victory over Iowa. OSU got a lot of computer credits for that game, based on the prior strength of Iowa's victories with a seasoned Senior QB, even though they only had to play against a RS Freshman leading the Iowa offense. Billingsly, though, gives OSU credit (perhaps even extra credit) for beating a team with a high rating at the time of the game. Yet, in this caes, that rating was no longer indicative of Iowa's strength.
I am not sure what to do about this problem. I suppose you could consider starters lost to injury at the time of the game. But the OSU MI game in Hart and Henne's last year would show that both played that game, even though Hart was running on one leg and Henne playing with one arm (with his throwing arm severely injured). So, I guess that any attempt to consider injuries could mislead and, in any case, would get far too complicated.
That probably brings us back to using the end-of-season ratings. In any case, the proof of the worth of a system lies in its ability to predict the actual outcome of games. I think that the prediction abilities of the existing BCS computer rankings should be tested and compared. Then the best performing one (or combination) could replace the current system.
The injury problem is one that I don't think a computer ranking can really handle (at least, not without making it absurdly complex and doing individual player ratings). We probably have to simply cope with the fact that the development of teams (for better or worse) throughout the season is going lead to some opponents being overvalued or undervalued relative to their real strength at game time (as an extreme example, teams that lost to Dennis Dixon-led Oregon a couple years ago won't get as much credit as they should for a tough game thanks to the later exploits of Brady Leaf-led Oregon, and teams that beat Brady Leaf-led Oregon will get more credit than they deserve due to the earlier exploits of Dennis Dixon-led Oregon).
As for testing the "prediction abilities" of the existing rankings ... how exactly would you do that? There's not enough data to get a good base on which to build predictions before the end of the season and roster turnover wipes the slate clean. Going back to retroactively "predict" (that is, see what percentage of games were won by the higher-ranked team, and whether this increases with the gap in the rankings) might be workable, but I'm not convinced it really tells us anything. And looking at the end-of-season polls as a comparison is pointless because we're trying to eliminate the polls, not ape them (this means you, Billingsley!).